Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Beyond Last-Click: An Optimal Mechanism for Ad Attribution

Nan An, Weian Li, Qi Qi, Changyuan Yu, Liang Zhang

TL;DR

The paper reframes advertising attribution as a mechanism-design problem where platforms may strategically delay reports. It shows Last-Click is not DSIC and introduces the Peer-Validated Mechanism (PVM), which is DSIC and achieves superior accuracy and fairness, with optimality in homogeneous settings and provable bounds in heterogeneous ones. The authors provide theoretical guarantees and demonstrate, via numerical experiments with real-world-like data, that PVM outperforms Last-Click in both accuracy and fairness. The work offers a principled, incentive-compatible foundation for robust attribution in multi-platform advertising ecosystems and suggests avenues for further refinements under distributional correlations.

Abstract

Accurate attribution for multiple platforms is critical for evaluating performance-based advertising. However, existing attribution methods rely heavily on the heuristic methods, e.g., Last-Click Mechanism (LCM) which always allocates the attribution to the platform with the latest report, lacking theoretical guarantees for attribution accuracy. In this work, we propose a novel theoretical model for the advertising attribution problem, in which we aim to design the optimal dominant strategy incentive compatible (DSIC) mechanisms and evaluate their performance. We first show that LCM is not DSIC and performs poorly in terms of accuracy and fairness. To address this limitation, we introduce the Peer-Validated Mechanism (PVM), a DSIC mechanism in which a platform's attribution depends solely on the reports of other platforms. We then examine the accuracy of PVM across both homogeneous and heterogeneous settings, and provide provable accuracy bounds for each case. Notably, we show that PVM is the optimal DSIC mechanism in the homogeneous setting. Finally, numerical experiments are conducted to show that PVM consistently outperforms LCM in terms of attribution accuracy and fairness.

Beyond Last-Click: An Optimal Mechanism for Ad Attribution

TL;DR

The paper reframes advertising attribution as a mechanism-design problem where platforms may strategically delay reports. It shows Last-Click is not DSIC and introduces the Peer-Validated Mechanism (PVM), which is DSIC and achieves superior accuracy and fairness, with optimality in homogeneous settings and provable bounds in heterogeneous ones. The authors provide theoretical guarantees and demonstrate, via numerical experiments with real-world-like data, that PVM outperforms Last-Click in both accuracy and fairness. The work offers a principled, incentive-compatible foundation for robust attribution in multi-platform advertising ecosystems and suggests avenues for further refinements under distributional correlations.

Abstract

Accurate attribution for multiple platforms is critical for evaluating performance-based advertising. However, existing attribution methods rely heavily on the heuristic methods, e.g., Last-Click Mechanism (LCM) which always allocates the attribution to the platform with the latest report, lacking theoretical guarantees for attribution accuracy. In this work, we propose a novel theoretical model for the advertising attribution problem, in which we aim to design the optimal dominant strategy incentive compatible (DSIC) mechanisms and evaluate their performance. We first show that LCM is not DSIC and performs poorly in terms of accuracy and fairness. To address this limitation, we introduce the Peer-Validated Mechanism (PVM), a DSIC mechanism in which a platform's attribution depends solely on the reports of other platforms. We then examine the accuracy of PVM across both homogeneous and heterogeneous settings, and provide provable accuracy bounds for each case. Notably, we show that PVM is the optimal DSIC mechanism in the homogeneous setting. Finally, numerical experiments are conducted to show that PVM consistently outperforms LCM in terms of attribution accuracy and fairness.

Paper Structure

This paper contains 53 sections, 15 theorems, 156 equations, 2 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Main Contribution
  3. Related Work
  4. Model and Preliminaries
  5. Real-World Advertising Scenario
  6. Advertising Attribution Model
  7. Attribution Mechanism
  8. The Last-Click Mechanism
  9. Accuracy Analysis
  10. The Peer-Validated Mechanism
  11. Optimality of PVM for Homogeneous Platforms
  12. Accuracy Analysis
  13. Fairness of PVM
  14. Numerical Experiments
  15. Conclusion and Discussion
  16. ...and 38 more sections

Key Result

Theorem 1

An attribution mechanism $\mathcal{M}$ satisfies DSIC if and only if, for every platform $i$ and any fixed reports from other platforms $\boldsymbol{r}_{-i}$, the credit $x_i(r_i,\boldsymbol{r}_{-i})$ is non-increasing in its own report $r_{i}$.

Figures (2)

  • Figure 1: Conversion-aligned time transformation
  • Figure 2: Normalized PDFs for 4 platforms

Theorems & Definitions (34)

  • Definition 1: DSIC
  • Theorem 1
  • Definition 2: Last-Click Mechanism
  • Proposition 1
  • Theorem 2
  • Theorem 3
  • Theorem 4
  • Definition 3: Peer-Validated Mechanism
  • Proposition 2
  • Theorem 5
  • ...and 24 more